Video Automation (or Programmatic)



The term “programmatic” has become ambiguous shorthand for some or all of a diverse range of platforms, tools, and processes in digital advertising. Despite its widespread use, there is still confusion around how to differentiate one programmatic solution from another, specifically as it relates to digital video.

The IAB Data Center of Excellence has specifically acknowledged the functional limitations of the word “programmatic” to meaningfully describe software-driven processes, formats and tasks, and has subsequently developed a Framework for Advertising Automation that encourages the use of more nuanced language when discussing this class of tools and capabilities. It does so by focusing readers on specific supply-chain processes and associated sub-tasks that more specifically illustrate the utility that data and software add to media planning and selling.

IAB defines programmatic very simply as the automated buying and selling of inventory. Thinking about automation beyond simple “yes” or “no” dichotomies, the reality is that there are many distinct processes within buying and selling workflows that can and cannot be automated, including targeting, forecasting, transacting, delivery, and reporting. The degree to which these processes can be automated is largely a function of the channels within which the video inventory is being sourced— desktop, mobile, OTT, linear TV—as well as the technical infrastructures upon which they’re being monetized—reserved inventory prioritization within an ad server, network-based monetization via tags or SDKs or exchange-based monetization. 

In the IAB report, Programmatic Video: a Spectrum of Automation, each process is outlined along with channels and technical monetization infrastructures in a grid that illustrates range of automation capabilities by platform, highlighting the degree of automation in each step of the supply chain.

This section of the guide expands on the concepts described in the IAB Programmatic Video Paper.

Now that automation via software and data has become the de-facto means of executing digital advertising investments—just as automation has become central and disruptive to most industries—understanding and evolving the roles and utility of each component involved in automation is critical to ensuring an effective marketplace.

Video Automation Opportunity 

According to the 2016 IAB NewFronts Video Ad Spend Report, programmatic video buying is seeing broad adoption and steady growth, accounting for 45 percent of all Digital Video dollars spent and is expected to continue to grow.

Programmatic video, which eMarketer estimates will account for 69 percent of U.S. digital video ad spending in 2017, is mutating into a more holistic endeavor. The expectation is the term “programmatic” may eventually fall away as automation and convergence of digital video and linear TV solidify to become the new industry norm. 

According to the IAB 2017 Video Ad Spend Study, multiple factors will drive automated buying and selling to account for nearly three-quarters of U.S. video ad spending by 2018 including:

  • Greater comfort with automated tools and audience-based selling will drive publishers to offer a greater portion of their inventory for purchase in an automated way as advertisers are willing to pay more to reach the right audiences.

  • More demand for audience-informed video ad buys will drive publishers to adopt automation to enable such data-driven capabilities.

  • Improved automated ad technologies offering guarantees both private and similar to traditional direct-sold agreements, will provide greater comfort for publishers as they push their most precious asset through the automated pipes.

Automated video bidding was driven by the need to revolutionize the waterfall model[37] and democratize the access to a publisher’s video inventory. For publishers, it offers a way to connect and manage multiple demand partners easily, and gain more flexibility and control over the transaction process. For brands, though the ability to buy video in an automated fashion exists today, doing it at scale with precision targeting is challenging due to the scarcity of premium video content.

IAB hosts a deep-dive crash course on programmatic advertising and inventory management called Programmatic 360: Automation Decoded.

Ad Technology vs. Publisher Revenues

While ad technology fees incurred by buyers and sellers vary greatly depending on which services are applied and the levels of managed services being requested, they cumulatively represent a significant portion of programmatic (automated buying / selling) revenues. IAB estimates that, in 2014, ad technology services cumulatively captured 55 percent of programmatic (automated buying / selling) revenue while publishers captured the remaining 45 percent.[38]

While these fees are an inevitable and important part of the ecosystem that support the platforms that enable automated transactions, it’s clear that buyers and sellers need to have better tools to help them evaluate the overall costs of automated technologies and services relative to the benefits they provide. For those who want to go one step further, IAB developed a programmatic fee transparency calculator. The calculator is intended to provide advertisers and publishers more granular, campaign specific cost breakouts based on their particular implementations. The tool tabulates the overall cost of each actor in the supply chain (as a percentage of the effective CPM) after the user enters his/her channel specific planning rates, budgets, and ad technologies/services.

Video Automation Landscape 

The diversity of automated buying / selling ad technology services—as well as their associated implementation options and cost models—has exploded over the past several years as the industry responds to changing marketplace dynamics. The increasing complexity in the programmatic value chain has also created unforeseen consequences. 

First, given the number of ad tech services commonly applied to automated buys, it has become difficult for advertisers and marketers to keep track of what specific service types are being used, and their relative value compared to other components of the buy. Second, there is a lack of communication between buyers and sellers about the ad technologies being used by each side, as well as the fees that are removed in the bidding process[39] before publishers’ net CPMs are realized. This creates a discrepancy between buy and sell side inventory valuations, which has the potential to erode trust in the marketplace and reduce inventory and price liquidity. The first step in improving transparency is to deconstruct the individual components of these ad technology layers and their value propositions.

While it would be self-defeating to identify and define every type of automated service in a constantly evolving landscape, the table describes the most common automated ad technologies and services within the supply chain, frequently used cost models, and the party (buy vs. sell) that generally applies the service.

This graph shows an example of some of the participants that might be involved in the automated media buying process.

Participants, Roles, and Monetization Infrastructure 

This section explores the core ad technology—the cost of doing business—including Supply-Side Platforms (SSPs), Demand-Side Platforms (DSPs), and their monetization infrastructures. 

SSPs (Supply Side Platforms) 

In addressing SSPs, we’ll examine the various ways in which an SSP can make calls to demand partners and the ways in which an SSP integrates demand partners into their auction. We’ll also review potential errors that can be encountered when integrating demand sources as well as the various tactics a publisher, using an SSP, can implement to push their inventory through the programmatic pipes. We’ll finish this section with a game to test your knowledge.

Asynchronous calls mean everything (objects, nodes, signals) happens in parallel and the SSP receives the responses typically before the page has loaded. This is the way most SSP auctions work these days. The bid request is ‘broadcasted’ to all buyers at the same time which is the most ef fiient way of calling video buyers as it reduces latency and it also creates competition.

Synchronous calls wait until the previous object has loaded before starting the next object. Its architecture stems from the traditional ‘waterfall,’ which causes additional latency between page load and ad delivery. 

Integrations refer to the process of adding a demand source to an exchange environment. It can also refer to the process of linking a video player to an ad server. In this section, we will go through the two primary ways in which an SSP adds demand sources to their auction tag based/fixed demand tags, and server-to-server integrations.

Tag Based/Fixed Demand Tags

Fixed demand tags are typically used as a way to compensate for low server-to-server demand (the result of few integrated DSPs) or to increase bid density. Ad networks that do not adhere to the RTB protocol will send fixed demand tags at specific prices. Let’s illustrate this with an example with a fixed tag of $15.

A publisher makes video impressions available through an SSP. The bid request goes out and demand partners send bid responses. DSP responses are in real time, but ad networks respond with a fixed demand tag (SSPs enable this via proxy bids). These tags are usually VPAID tags, as they allow the demand partner to perform ad decisioning and cookie collection. When they respond with a fixed demand tag, in parallel, they initiate the process of finding a buyer; if they can’t find one, they respond with an empty VAST tag. By the time this happens, other demand sources might find that $15 is too high of a price and assume that someone else bid and won the impression.

The end result is a wasted impression with an inflated price that never clears the auction, and a poor user experience.

Server-to-Server Integrations

Server-to-server integrations are designed to enable faster transfer of data between organizations, quick activation, and enhanced reporting.

In a server-to-server integration, the ad platform is first called from the “client,” which is typically a browser, app, or video player running on a user’s device. After that, the client only waits for the final creative—to show the user—with all decisioning logic and communication with buyers, running through data centers that are optimized for these direct connections. These connections are typically very fast, they may be in the same, or geographically close data centers, and usually do not need to navigate typical web traffic. They are also fast because they are not limited by the speed of the user’s device but can be optimized by the ad tech vendor’s server farms.

Because the buyer does not have a direct connection to the client (browser, app or video player), it doesn’t have access to their cookie, so it may be more difficult for them to identify the user. One method to bring this data to the buyer, is for the buying and selling platforms to “synch” cookies. This is an ongoing process that allows the seller to share a user identifier with the buyer. In some cases (such as mobile in-app supply) the client has a device ID that can be passed along via parameters to each party.

To learn more about mobile identity, please visit the IAB “Mobile Identity: A Best Practices Primer for Mobile & Cross-Platform Device Marketing.” 

What Could Go Wrong: Causes for Wasted Impressions

Broadly speaking, there are two reasons for a publisher to waste a valuable video impression: an empty VAST tag, or a VPAID error. These errors lead to latency, wasted impressions, breaking competition and poor user experience.

When a demand partner responds with a VAST tag that has a VPAID component in it, the video player may think that because the scripting language started running (the demand partner responds with a VPAID component), an ad will be served. However, this is a bad assumption. When the VPAID logic executes, it may make calls to different demand partners to see if they return an ad. While this was

not the original VPAID goal, the technology has been used innovatively (or hijacked from a different perspective) to perform these tasks.

For VPAID creatives, only an “AdImpression” VPAID event sent to the player should be used by a player to indicate that a VPAID ad is playing.

When a DSP uses VPAID opportunistically to do client-side bidding/decisioning like this, the player only ever sees the initial VAST response that contains the VPAID bidding wrapper creative. The player loads this wrapper and allows the VPAID code to execute. After this, any VAST ad that is loaded is done so by the VPAID code using its own loader and the player isn’t even aware of it. The way VPAID signals to the player that there was no creative, is by calling the “AdError” VPAID method. This returns control back to the player and the player can then decide to try another demand source or give up. Some video players are sophisticated enough to understand there’s no creative and thus the player will call on a different ad source.

Another reason for wasted impressions is when a VPAID wrapper loads another VPAID wrapper in itself. If there is a VPAID error, this is communicated to the parent VPAID wrapper which may then call on another demand partner. All this back and forth adds latency, which is the number one reason for a user to abort their session or close the video player, which results in a wasted impression. Additionally, if the ad doesn’t show up on time, the publisher needs to give up the ad impression and show the content instead, therefore also wasting an impression. 

DSP (Demand Side Platforms)/Exchange Integrations 

Integrations, What Could Go Wrong?

Now that we have gone through the SSP set ups, tactics, best practices and priorities, let’s look at the various scenarios and possible errors from the DSPs side.

There are several errors we can encounter while integrating a DSP for video. It is worth mentioning that if a DSP has integrated for display and they now want to buy video impressions, they will have to go through a new RTB integration for video. This requires technical expertise on both sides and it usually needs to be in a DSPs integration queue, which can take one year or more. If you are planning on integrating DSPs, plan ahead and target video DSPs first. The open RTB integration can take between 4-12 weeks based on resources on both sides.

The errors you might encounter include:

  • Not connecting the data center correctly

  • Not seeing impressions

  • Not responding to the ad call in the allotted time (usually less than 100 mms)

  • Seeing errors in the bid request

The technical account manager on the SSP side will check for all of these possible errors in a phased approach following the IAB Open RTB protocol

General or Specific?

Do you go for a generic or a video only DSP? It will all depend on your business, whether you want unified reporting and how your video spend compares to the rest of the formats. If that number is high, there will be advantages in dealing with a video specialist as they are created with the eventual convergence of video with TV in mind and are designed to drive metrics important to TV-centric brand advertisers. Generally, programmatic video and TV campaigns that drive the greatest brand results follow a set of principles quite different from those of display, including:

  • A combination of RTB & reserved buying and forecasting to ensure campaign certainty in a scarce marketplace

  • Data fluidity across time and devices to manage frequency and drive relevancy

  • Putting importance on controlling viewability, fraud, and favorable brand environments

  • Striving for fewer but better-targeted ads to reach the right consumers, thus eliminating waste and preserving user experience

  • Optimization to an advertiser’s true key performance indicators (KPIs)

Managed vs Self Service

In the ad-tech world, solutions/software can be categorized as either self-service or managed service. The distinction here is as it sounds—self-service tools can be accessed and used directly by the client and independent of the vendor, while managed-service tools may have a component that can be used without help, but requires someone from the vendor to implement actions. Managed services involve setting up, managing, and optimizing automated delivery, as well as building and maintaining buy- side infrastructure such as inventory, technology, data, and reporting relationships.

Due to advertisers’ varying degrees of fluency with automated buying tools and concepts, there has been a stratification of managed service options that vary significantly in terms of service and fee levels. For a list of the most common buy-side managed service options used, please refer to the IAB “The Programmatic Supply Chain: Deconstructing the Anatomy of a Programmatic CPM.” 

Monetization Infrastructure 

There are two fundamental ways to buy and sell media: dealing directly with the marketer, or indirectly via a technology platform that would allow the publisher to make their inventory available in some automated way. Historically, video has been transacted in a direct way. Direct selling of video inventory remains the standard approach to doing business. For a deeper dive on the reasoning behind method visit the IAB 2016 white paper, Programmatic Video: A Spectrum of Automation.

In the TV business, each year there are flashy events called upfronts where publishers and marketers commit to buying media in advance. Think about the Super Bowl; the ads we see during this big event were negotiated the prior year.

Video and TV are so constrained for supply that marketers want to ensure they have the spot available to capture their target audience. Though the concept is the same—securing or guaranteeing certain amount of inventory—the execution has evolved due to the technological advancements that have allowed the industry to apply automation and more data-driven buys.

The following table shows some of the ways in which “programmatic” was being applied in 2013. Fast forward to 2017 and though there are more variations, the principle remains the same, there can be automation of the transaction (Real-Time Biding = RTB), of the process (Automated Guaranteed = AG) or a combination of both. 

Inventory Hierarchy

Putting all the pieces together in a pyramid, this is how a typical inventory hierarchy looks like.

At the very top of the pyramid, we have direct sold inventory which is guaranteed, as we move down the pyramid, we move towards RTB or real time decisioning (RTD) environments where impressions can or cannot be guaranteed. 

Open Exchange/Auctions

Per IAB Programmatic Video: A Spectrum of Automation, IAB defines exchanges as exchange-based technology solutions like DSPs and SSPs, that facilitate workflow automation, enabling buyers and sellers to scale the number of transactions to levels beyond what is possible with network or direct buys. This automation is made possible through mass-market software tools: demand-side platforms on the buy side (DSPs) and supply-side platforms (SSPs) on the sell side. These tools are generally made available to users on a self-service basis; meaning campaign structure, pacing, and optimization are done directly by the user instead of network intermediaries.

A publisher will generally allow any and all buyers to participate in accessing their inventory. Usually there is no direct relationship with the buyer.

  • Publishers may choose to use block lists to prevent a certain advertiser/brand from buying their inventory (a publisher might have a direct agreement with the said brand, hence wants to prevent them from buying the same inventory) and floor prices to ensure rate cards are adhered.

  • Advertisers usually buy inventory in the open market when trying to buy a broad demographic, where a sniper approach might not be necessary. In the display world, where there’s unlimited supply of impressions an advertiser is often unaware of what publisher they are buying on, DSPs usually present a list of exchanges/SSPs to the buyer that they automatically opt into. Buyers may not know they are buying a publisher’s inventory. Because of this, publishers can participate in the open auction on a blind basis. 

In video, where CPMs are much higher than in display, publishers must show some transparency or the likelihood of buyers buying their inventory will be low.

The Open Exchange runs through Real-Time Bidding (RTB) protocol which has been created by IAB tech lab, the latest release is of Open RTB2.5 and offers the ability for buyers to decision down to the impression level. IAB tech lab members interested in participation in the RTB protocol update working group may contact [email protected].

Buyers are able to buy video inventory in the Open Exchange today across multiple exchanges across both desktop and mobile (In App and Mobile Web) through the majority of demand-side platforms.

The main difference between open and private actions is that in the open auction, the publisher doesn’t know who is buying their inventory until it happens. They can implement certain tactics to control their inventory, but these are not bulletproofed.

Invitation-Only Auctions, Private Exchanges/PMPs, and Deal ID

An “invitation-only auction,” or a “private auction” is a type of auction where a publisher restricts participation to select buyers/advertisers via whitelists and block lists. A publisher may choose to not participate in an open auction and only run an invitation-only auction. It is important to note that in an invitation-only auction, buyers will be expected to bid on inventory. A publisher may choose to expose different information such as transparency or data, through the use of Deal IDs or Line Items to add value to this select group of buyers while participating in this tactic.

A “private exchange” is characterized by both one-to-one and one-to-many transactions. A private marketplace (PMP) is usually operated by one (or a few) large enterprise(s) and is open to the enterprises’ strategic trading partners along its entire supply chain. For sellers, the advantage of entering into a private exchange for video offers greater control, enabling access to their premium buying platforms and more brand safety controls to the tens of thousands of advertisers they serve. For buyers, it provides greater access over placement, viewability and first-party audience data/segments.

A “Deal ID” is a unique identifier for a specific private marketplace deal between buyer(s) and seller(s). Think of a Deal ID as a feature of a PMP. Deal IDs contain a unique string token that is passed on the bid call as a targetable unit between buyer and seller (DSP and SSP). Impressions are classified into a Deal ID based upon selling rules. Bids are then generated based on the set of rules that align to that particular deal. Deal IDs have evolved greatly since their inception to include multiple buyers and sellers.

As an example, if AMNET (trading desk) wanted to execute a video PMP deal on behalf of three of their brands they can execute that deal through one Deal ID, they would set the Deal ID through one of their DSP partners. That same Deal ID can be used across multiple sellers/SSP. The unique token only relates and can communicate between one DSP and one SSP. The evolution in Deal ID set up greatly benefits video, provides greater bid density and competition for the seller, and greater reach and frequency for the buyer.

Automated Guaranteed (AG)

This type of transaction most closely mirrors a traditional digital direct sale. AG helps publishers to manage direct sold, guaranteed deals—Insertion Orders, or IOs—in an automated way. Publishers make their inventory available through their ad server by grouping inventory based on advanced targeting.

The deal is negotiated directly between buyer and seller, the inventory and pricing are guaranteed, and the campaign runs at the same priority as other direct deals in the ad server. They then set prices for each audience package and make it discoverable to buyers through an interface or API connection to the buyer platform. The buyer can then pick the number of impressions that they want to buy or forward reserve, negotiate price and launch the campaign. This eliminates the need for inventory discovery calls as well as any on-going back-and-forth communication which takes a lot of manual efforts and time. The publisher and buyer get engaged directly which brings maximum transparency.

The programmatic element of the transaction that differentiates it from a traditional direct sale is the automation of the RFP and campaign trafficking process. Negotiation through to fulfillment can be, should the publisher desire, completed within the technology platform providing the automated reserve functionality.

AG runs via the OpenDirect protocol specification; the latest update, Open Direct 1.5, was released in September 2016. The IAB OpenDirect protocol is a powerful API designed to support automated guaranteed business models and marketplaces, which originally were introduced to automate and replace the traditional IO process and which now have far-reaching applications.

IAB is constantly updating its protocols; if you are an IAB Tech Lab member interested in participation in the OpenDirect protocol update working group, please contact [email protected].

Setups, Tactics, and Best Practices 

Once publishers and marketers understand the differences between the open market, private marketplaces and automated guaranteed setup, they can think of tactics to implement. 

Open vs. Private Marketplaces (PMP) 

Open marketplaces are exchanges that aggregate requests from many different publishers in one general market. This allows a lot of advantages over private marketplaces, which are often limited to a single supply source. 

Benefits of the Open Auction for Publishers 

Publishers pushing inventory in the open market can access valuable insights, such as:

  • Interested buyers can be great lead sources for their direct sales team.

  • Blocked buyers can be great leads for Deal IDs.

  • Buyers buying a lot of inventory in a certain audience, can lead publisher to create a custom audience package (sold at a higher CPM).

This is where the SSP acts as an extension of a publisher’s sales team providing their direct sales team with real-time data to upsell into their clients and sometimes new leads. 

Benefits of the Open Auction for Advertisers 

Buying inventory in the open market provides DSPs with scale and data. DSPs have the capability
of processing many campaigns at the same time. Having the information passed in the bid requests coupled with a robust and compliant RTB integration will enable them to source the right inventory to deliver against their campaigns. 

Private Marketplaces 

According to the IAB blog post: ‘To Private Marketplace or Not to Private Marketplace - That is the Question...’ for a while, it seemed that Private Marketplaces (PMPs) were the solution to every problem. Worried about the quality of inventory in the open auction? Worried about automation (or ‘programmatic’) becoming a race to the bottom? Worried about control as a publisher? PMPs were the answer.

In the rush to set up a Private Marketplace, far too few people were evaluating whether a Private Marketplace was truly the most appropriate approach. Even fewer were assessing ahead of time what the overlap was between the buyer’s target audience and the publisher’s audience. The result was PMPs not delivering ROI and the volume of transactions through PMPs not meeting expectations.

The answer: a PMP checklist. The aim of the checklist is to ensure that buyers and sellers are on the same page about what they are trying to achieve from their private marketplace and that they can appropriately assess whether it is the right channel through which to transact. The PMP checklist provides a list of issues that buyers and sellers need to discuss and agree to ensure ROI from their private marketplaces. 

The private marketplace checklist is divided into three sections:

1. Consideration: The aim at this stage is to determine if a private marketplace is the right approach and will yield positive ROI
by comparing the buyer’s needs and target audience with publisher’s capabilities and audience.

2. Activation: Having established that a private marketplace is the right approach, the next stage seeks to ensure that buyer and seller agree on parties involved, inventory transparency, and financial terms/timing to deliver ROI.

3. Troubleshooting: The aim of the final stage is to help identify common issues such as low impression volume, poor win rate, and flighting/targeting that may arise once the private marketplace is set up.

Private marketplaces are different as they are typically a one-to-one relationship between sellers and buyers. The publisher knows who is buying the inventory. Deals are negotiated beforehand.

Private marketplaces allow publishers to keep tight control, only allowing certain trusted buyers to access their premium inventory. Typically, a publisher will retain complete control over the operation. In addition, this approach still allows them to route traffic to the open marketplace if the inventory isn’t cleared in the private environment.

Private marketplaces aren’t usually guaranteed, that is, the publisher agrees with the buyer to push a certain audience, and the buyer has the first right of refusal at the agreed price. In certain instances, the private marketplace can be guaranteed which means that the publisher will have to push targeted audience impressions until the buyer has found/matched all the impressions for the audience they have targeted. These types of deals are usually at a higher CPM to compensate for wasted impressions.

It can at times be a good strategy for marketers and publishers to use both open and private marketplaces. The publisher wants to push the most premium impressions to identified buyers, and if these aren’t bought in the private exchange, send them to the open exchange with a floor price and certain restrictions so that it doesn’t send the wrong signal to the market (that buying in an automated way = cheap inventory). From a buyer’s perspective, they can buy the bulk of inventory in the open market, and set up private marketplaces for distinct advanced audience buying tactics.

Mass Reach vs. Precision Targeting 

The mass reach vs precision targeting tactic is one that is applied by the buy side.

Choosing the right tactic depends on the campaign’s goal. The mass reach versus precision targeting campaign looking for general awareness. Advertisers can do this with no targeting or by simply focusing on demo/lifestyle segments. If the brand is looking for something more precise or further down the funnel, then the precision targeting or “sniper” approach works better, ensuring the audience is more closely aligned with the target audience. Additionally, segments such as purchase-intent would work well here.

In the video world where premium impressions are scarce and CPM are so much higher than display, DSPs tend to use a more precise targeted approach, or sniper. 

Optimizing Deal ID 

Optimizing Deal IDs means monitoring deal health, audience match, and fill rates.

Optimization occurs on the DSP/SSP side by looking at metrics, audience segmentation alignment and bidding behaviors, along with white- and black-listing i.e. if the Publisher is using Krux audience segments, while the advertiser is using Lotame, there will be a discrepancy in what is classified as a sports enthusiast in both DMPs.

As publishers are dealing with various SSPs and DSPs, there needs to be Deal ID maintenance. These need to be constantly monitored to check whether the DSP has ‘turned them on’ or whether the DSP is seeing impressions. Sometimes DSPs create evergreen deals that have no activity. It is also worth mentioning that each SSP will have its own Deal ID identifier, there isn’t a set nomenclature to call the Deal IDs, which adds one more layer of complexity and possible errors.

Please refer to the IAB PMP Checklist for additional considerations, activation, and troubleshooting of PMP deals.

Evergreen Deals 

These deals are created to utilize one Deal ID across an entire DSP, trading desk, or agency, they are usually open ended and the targeting criteria doesn’t change. It’s like setting a “spray and pray” tactic in a private marketplace environment for a set audience that is open indefinitely. These are good deals to have, however one should monitor for activity, as if they are in hiatus, they should be closed. 

Priority Level Games: Who Wins? 

Ad servers and SSPs have the ability to set priorities in the form of inventory, pricing, or targeting, to name a few. An ad server might have as its first priority to sell all of the direct sold inventory first, or to push a Deal ID to priority one if the direct deals are pacing ahead of schedule. When a publisher decides to move some of their inventory to an auction type setup, they use a supply side platform (SSP) that enables them to push through the impressions to the marketplace. SSPs also have the ability to set priorities, just like servers. In order to understand priority levels and who wins based on a scenario, let’s play a game. 

Game #1: We have two scenarios. In the first the floor price is $10, so unless buyers return a response with a bid of at least $10.01, they won’t be considered a participant in the auction. We have two responses above the floor price. Who wins, and at what price?

And The Winner is? 

  1. DSP1 wins as they have Priority 1, but the clearing price is $12 because of the 2nd price auction principle.

  2. Even though DSP2 bid more, DSP1 had priority 1, which trumps the higher bid response. 

And The Winner is?

Game #2: Eight potential bid responses above the floor price, though #5 chose not to respond. Who is the winner?

  1. This looks complex, but focus on the priority. In this case, DSP1 wins and the clearing price is the one they made. 

And The Winner is?

Game #3: In the first scenario, we have three possible demand partners responding to a bid request, two of them respond above the floor price.

  1. DSP1 wins because they have priority 1 and the Deal ID with first look didn’t bid, the winning bid goes to DSP1.

  2. In this case, the Deal ID has first look, as there aren’t any other bids, the clearing price is the floor + one cent. 

Header Bidding 

“Header bidding, also known as pre-bidding or advanced bidding is an advanced automated technique wherein publishers offer inventory to multiple ad partners simultaneously before making calls to their ad servers. By letting multiple demand sources bid on the same inventory at the same time (as opposed to sequentially), publishers increase their advertising yield and revenue.”[40]

Header bidding is a process that allows publishers to auction ad impressions in a flat parallel auction, across multiple sources of demand facilitated by a snippet of JavaScript in the header of a web page.[41]

In most publisher / ad server set ups, demand sources are checked sequentially or in a ‘waterfall’; direct sold inventory is prioritized over ad networks and indirect demand sources. Header bidding auctions occur before the ad server is called; demand partners respond with bids, which are then passed to the ad server where the winning advertiser is determined. 

Header Bidding for Video Advertising 

Because of its success in display, publishers have begun to explore header bidding in video.

Today, header bidding is more widely used in display than in video. Display advertising is traditionally more exchange-centric/RTB-driven, than video. However, publishers understand the need to employ data-driven, audience-based selling strategies and the push to automate some of the processes. As such, they are eager to explore other strategies but want to ensure they protect their scarce premium video inventory as it becomes available through automated channels/mediums.

In video, the ad call comes from a video player, not from a page header. Therefore, video ad calls have different requirements than the display world. In other words, with video, there is an entire additional layer of technology that needs to be incorporated into the site’s ad stack, so the code used in display header bidding needs to be part of the video player, rather than the header itself. The video player then needs to communicate its decision about where to send the ad to the ad server. And while this communication works with some video ad servers, it does not work for others. 

Through some workarounds, a version of “header bidding” is possible in video advertising through the video player.

  • The benefits of using header bidding for video would be similar to those in display advertising, e.g. the ability to have multiple demand sources compete for the same available impression, thus theoretically increasing overall yield for the publisher.

  • In response to concerns about speed, there is growing traction around the idea of server-side header bidding, which is faster than traditional header bidding because it happens on a remote server, rather than the user’s browser. This faster speed means publishers can collect the widest range of available bidding sources before shortlisting the optimum set and sending it back to the client for action, thus increasing yield.

Automation of the TV Buying and Selling Process 

The total TV advertising market size is about $200 billion worldwide. For the U.S., the overall TV market size currently is about $70 billion, with Magna Global estimating $10 billion for Programmatic TV (PTV) advertising by 2019

Overview: What is TV Automation (or Programmatic TV)? 

The use of the term programmatic or automation in the context of television is challenged and confusing, largely because the current capabilities in TV are not comparable relative to programmatic in display media bought and sold in an automated way. For this reason there are those who prefer to use the term automated, or audience index-based TV buying. IDC defines Programmatic TV or Automated TV as: The use of software platforms to automate the workflow of TV advertising and to improve the effectiveness of TV advertising through the use of advanced TV audience data. Platforms do not need to support auction based sales or real-time bidding in order to qualify as PTV.[42]

  • Programmatic TV or Automates TV automates the workflow of advertising on linear TV such as campaign planning, requests for proposals, price negotiation, copy management, scheduling ad insertions, the actual insertion of copy, reporting, billing, etc.

  • Programmatic TV or Automated TV uses a more data-driven approach beyond standard target demographics. It leverages set-top box data of millions of households, allowing more granular targeting. 

Potential Benefits of TV Automation 

With programmatic TV or Automated TV, marketers get better targeting, access to new inventory, and a unified interface integrating and simplifying the workflow. Programmatic TV or Automated TV helps sellers to improve sales:
  • By overlaying advanced audience data, their clients’ targeting capabilities are vastly improved, thereby making inventory more valuable.
  • Allows looking at inventory in smaller increments than in the traditional 30-minute-segment view, thereby unlocking undersold inventory.
  • More inventory is exposed to more buyers through tapping into the automated ecosystem. Multichannel video programming distributors (MVPDs) can now sell local advertising to the national market.
  • Programmatic TV or Automated TV reduces sellers’ costs through automating the workflow, which today is in large parts is still manual, and therefore slow, error-prone, and expensive. 

The New TV Buying / Selling Landscape: Who Are the Participants? 

There are multiple categories of players making up the TV buying/selling landscape including:

  1. The TV advertisers and media agencies who book campaigns.

  2. The demand side platforms (DSPs) who act as automated buying platforms mainly fulfilling two functions: Marrying data from DMPs with the available inventory and then executing the buy in an automated way.

  3. The sell or supply-side platforms (SSPs) who act as automated selling platforms. These platforms integrate into legacy TV systems for each network to make the automated buy technically possible.

  4. The data management platforms (DMPs) such as Rentrak, who layer audience data, engagement metrics, purchase data or other sources on top of the TV program to allow smart buying decisions.

  5. The publishers (national broadcasters, station groups, and cable TV networks) who supply the inventory. 

Automated Guaranteed 

Automated Guaranteed is an automated representation of the traditional Insertion Order (IO), and is comparable to traditional ad serving but through automated channels.

The buyer and seller agree on a deal ID (an identifier used to match buyers and sellers), which will represent the IO (with criteria such as minimum bid price, type of ad unit, site section etc.) in each of the automated platforms—the DSP for the buyer and the SSP for the seller. When this deal ID appears in the bid request, each side understands that this is the previously agreed on (a priori) deal. The primary components of an automated guaranteed deal are a fixed CPM rate for that supply, a timeframe it should run, a revenue or impression goal for the campaign, and the specific supply criteria of the IO.

In an automated guaranteed relationship, the seller has committed to giving the buyer a certain amount of relevant inventory at the fixed rate, and is responsible for ensuring targeting is in place to satisfy the goals of the IO. The SSP is used to pace the automated deal, sending approved supply to the DSP for purchase, with the proper deal ID and CPM.

The buyer may use their own metrics to ensure the IO is satisfied, such as confirming geo-targeting via reporting, and making sure the campaign runs inside the proper time period. However, when the DSP is presented with an automated guaranteed deal ID they are required to fulfill their commitment by buying it at the given rate. They should expect that this inventory has been put aside for them and the publisher will not offer it for sale again if the buyer chooses not to buy.

Automated Guaranteed is ideal for publishers who have supplemental information they do not share in an open market. This includes first-party user data, content categorization, behavioral data, or premium positioning of the ad. This mechanism allows the publisher to target information in their SSP and get a good price for valuable supply without exposing this (often proprietary) information in a non-guaranteed environment.

Automated Guaranteed is great for advertisers who are looking to lock in supply and are willing to commit to a certain level of spend at a fixed rate. This commitment may reduce risk of unsold inventory on the publisher’s side and encourage the deal ID to be prioritized, which often also increases the quality of the supply.

For occasions when the buyer wants to choose from pre-filtered supply, and have the option to buy a portion of it at a fixed rate, the better mechanism is an unreserved fixed rate deal. This works similarly to AG, but does not have the expectation that the buyer will purchase everything that is sent. In these cases, the publisher will share more supply for the buyer to sort through and the campaign will primarily be paced by the DSP. While it is still offered at a fixed rate, the amount of supply available is not guaranteed. 

Header Bidding: Benefits & Challenges for Publishers and Advertisers 

Benefits for Publishers

Drawbacks for Publishers

Benefits for Advertisers

Drawbacks for Advertisers

Header bidding has gained traction among publishers as it provides more control over how their traffic is monetized. It also improves the inefficient sequential auctions of the tag based waterfall. This process brings in real-time demand to compete in a publisher’s ad-server.

After such a landslide of adoption of header bidding for display advertising, adoption of header bidding in video advertising has been relatively slow. 

Obstacles for Adoption of Header Bidding in Video

  • The marketplace of high-value video inventory is in need of process improvement, but faces unique challenges due to a highly-fragmented ad stack, for which it’s been difficult to build the universal solutions necessary for creating a parallel auction.

  • In header bidding for display, you deal directly with the ad server. Header bidding for video introduces additional complexity by having a video player intermediate the request with the ad server. With the multiple players and ad servers in the ever-changing video ecosystem, video header bidding solutions are constantly evolving to adapt.

  • Video has long been plagued by latency issues, which many users assume is attributed to bandwidth and loading high-weight content. The reality is that much of that latency stems from the long-trailing waterfall effect caused by trying to find an ad to fill the high-CPM ad slot and the resultant passbacks.

  • With video header bidding, parallel requests are enabled, removing the need for passbacks which would improve latency. Many publishers manage latency by setting a time limit on bid responses from partners.

  • Despite the barrier to entry, many publishers are seeing incredible returns from increased price competition for their video inventory. Video header bidding is a tool that will allow media companies to realize the true price of their video inventory through increased competition and transparency, and is a monetization strategy that every video content provider needs to understand.

Header bidding allows all demand sources to compete directly for every impression, enabling publishers to realize the full value of automation. Header bidding—whether implemented directly on a page, app, or via a wrapper—provides publishers with a convenient way to manage their own ad exchange, including auction pricing management, reporting and analytics, latency controls, and more. As wrappers get more sophisticated, these individual exchanges will also expand their capabilities. 

Server-Side Video Header Bidding Challenges

These include the possibility of cookie loss if the SSP’s cookie IDs don’t match the DSP’s cookie IDs. More information can be found in the Videology Knowledge Lab’s Header Bidding paper

For a view on Hybrid Video Header Bidding, please visit

Why Turn to Automation? Decide for Automation? 

For buyers, Automated TV or Programmatic TV (PTV) offers the promise of better targeting, access to new inventory, and a unified interface, which integrates and simplifies the workflow.

For sellers, programmatic TV has the potential—for some forms of inventory—to increase revenues by selling inventory at higher prices, and by selling unsold or undersold inventory. This is accomplished by the application of advanced audience data. PTV integrates and automates the workflow for sellers as well, saving time and money. Local cable and broadcasters have the chance to unlock the national market through the use of PTV.

Streaming video, time-shifted viewing, video on demand, more cable channels and non-TV distractions such as mobile apps and games, eat away time consumers spend watching linear TV, as such media owners react with new tactics. MVPDs have launched TV Everywhere services, distributing their content online and on mobile devices based on subscriber user authentication.

There are also factors that hold TV executives back from adopting Automated TV. One concern is that automation will commoditize and thus decrease the overall value of media owners’ inventory. The fear of losing control over pricing and entrenched pricing models is real and in some cases well-founded. Legacy technology also presents challenges. Finally, there is also cultural inertia and vested business interests, which resist embracing new ways to do business.

But there is also one major difference between the TV industry and the digital industry that may slow adoption. Digital advertising is a buyers’ market: There is an essentially unlimited supply of inventory that far outstrips demand. This puts strong competitive pressure on digital publishers to please agencies and brands, prompting publishers to adopt automation.

TV advertising on the other hand is a sellers’ market: There is a limited supply of inventory, which is outstripped by demand. This puts media owners in the driver’s seat. As in the digital space TV buyers want automation, but there is less pressure on sellers to offer it. In the end, the market will decide automation’s fate in the converging video and TV marketplaces. 

1: The Nielsen Company, Monitor Plus (Standard Calendar, Total includes B2B, National Internet (Display only), FSI Coupons), Oct. 2016
2: IAB/PwC 2016 Internet Advertising Revenue Report
3: To view the Top 10 Video Content Properties by Unique Viewers, see comScore’s monthly release of their Video metrix data.
4: Business Insider: “The US Digital Media Ad Revenue Report: The path to $100 billion in annual revenue by 2021.”
5: IAB/PwC 2016 Ad Revenue Report
6: Business Insider: “5 video advertising trends that will change your business.”
7: PwC: “2017-2021 Global Entertainment and Media Outlook.”
8: Forrester Research, 2016
9: Pathak, S. (2017). “Why digital advertising is experimenting with blockchain.” Digiday, 4 Apr 2017
10: Gross Rating Point: Measured by the % of households that tune into to a particular show or network and have the opportunity to see an ad.
11: Reach: Represents the total number of people exposed to the media plan or ad over a certain time period, based on the total size of the target audience.
12: Frequency: Is a measure of media repetition.
13: IAB’s Digital Video In-Stream Ad Format Guideline.
14: IAB Deep-Dive on In-Feed Ad Units: A Supplement to the IAB Native Advertising Playbook
15: Herrmann, J. & Isaac, M. (2016). “The Online Video View: We Can Count It, but Can We Count on It?” The New York Times, 2 Oct 2016
16: Cloud Servers: A cloud server is a logical server that is built, hosted and delivered through a cloud computing platform over the internet. Cloud servers possess and exhibit similar capabilities and functionality to a typical server but are accessed remotely from a cloud service provider.
17: Syndication is a term that is used in both print and broadcast media. It indicates content that for instance is purchased for use by a local newspaper, TV, or radio station. It is not produced by the media company’s owner but through an outside source.
18: XML: Extensible Markup Language (XML) defines a set of rules for encoding documents in a format that is both human/machine-readable
19: CSS: is the language for describing the presentation of Web pages, including colors, layout, and fonts. It allows one to adapt the presentation to different types of devices, such as large screens, small screens, or printers. CSS is independent of HTML and can be used with any XML-based markup language
20: Audience segments are subsets of user data signifying specific facts, interests and other attributes. Audience segments, and the techniques
21: URI is a string of characters used to identify a resource. Such identification enables interaction with representations of the resource over a network using specific protocols. Schemes specifying a concrete syntax and associated protocols define each URI. The most common form of URI is the Uniform Resource Locator (URL), frequently referred to informally as a web address.
22: Making Measurement Make Sense (3MS) is a cross-industry initiative founded by the American Association of Advertising Agencies (4A’s), the Association of National Advertisers (ANA), and Interactive Advertising Bureau (IAB). The Media Rating Council (MRC), an independent body, is responsible for setting and implementing measurement standards.
23: IAS 2016 year-end survey results
24: An “affiliate” is an entity that controls, is controlled by, or is under common control with another entity. Control of an entity means that one entity (1) has significant common ownership or operational control over the other, or (2) can exercise a controlling influence over the management or policies of the other entity. In addition, for an entity to be under the control of another—and thus be treated as first party under these entity’s policies.Principles—that entity must adhere to online behavioral advertising policies that are not materially inconsistent with the other
25: Deloitte. “2015 Global Mobile Consumer Survey: US Edition. The rise of the always-connected consumer.”
26: Mobile Spearheads Digital Video Advertising’s Growth.” eMarketer, 22 Feb 2016.
27: comScore Inc., Nielsen, and ZenithOptimedia
28: Nielsen Q2 2016 Comparable Metrics report.
29: Q2 2016 Comparable Metrics report.
30: AARP.
31: Nielsen Q2 2016 Comparable Metrics report.
32: Mobile Spearheads Digital Video Advertising’s Growth.
33: ANA reports 7.2B lost in Ad fraud
34: Snapchat internal data
35: The continued evolution of enhanced mobile experiences that overlay digital information on top of the physical world
36: Duffy, F. (2017). “Turner Takes on eSports in a Big Way.” NCTA Platform, 6 Jan 2017.
37: The waterfall model is a sequential process in which progress is seen as flowing steadily downwards (like a waterfall) through the various phases, in this case, the potential demand sources or buyers
38: IAB Programmatic Revenue Report 2014 Results. July 2015
39: IAB Transparency is the Key to Programmatic Success
40: WTF is header bidding?
41: Thomvest ventures
42: Programmatic TV definition per IDC
43: Lotame Bridge the TV ad gap & PWC Media forecast 2015, agency reports, front row advisory analysis
44: eMarketer More OTT Time Means More Ad Time
45: eMarketer’s Connected TV and Over-the-Top Video: The Living Room’s Place in the US Digital Video Ecosystem report.
46: Frank N. Magid Associates study
47: comScore 2016 U.S. Cross Platform Future in Focus report.
48: Nielsen’s first-quarter 2016 Total Audience Report.
49: Adobe U.S. Digital Video Benchmark 2Q15; adobe primetime; TV connected devices=apple TV, Roku, gaming console, amazon fire TV,smart TV, other
50: Based on 2016 total viewers Broadcast - Source: Nielsen. Prime time total viewers, Live+7; Broadcast data: 12/28/15-12/4/16
51: Based on 2016 total viewers Cable - Source: Nielsen; Live+SD numbers from 12/28/2015-12/18/2016
52: Based on 2016 TV household coverage. Networks supplied coverage percentages, except for Create TV, which came from Across Platforms consltancy
53: IDC
54: Video Landscape Report
55: IAB TV 20/20 Webinar, 2016. Videology % estimates from Nielsen, eMarketer trend data; Time spent data calculated by Videology from Nielsen and KPCB data. Highest rated programs based on Nielsen A18-49 Live+7 data.
56: “2016 U.S. Cross-Platform Future in Focus,” comScore
57: IAB Digital Video Landscape report
58: Any Given Minute Comparable Metrics Report, VAB 2016
59: IAB: The Programmatic Supply Chain: Deconstructing the Anatomy of a Programmatic CPM)
60: Ghostery, Inc. Interview with Ghostery CEO Scott Meyer. IAB Annual Leadership Conference. January 26, 2016.
61: IAB/EY Study released on Dec-15. Estimates are for the U.S. Market only. Industry-wide collaboration under the auspices of TAG is needed in order to forestall these criminal activities